Download Innovation diffusion and new product growth models

Intern. J. of Research in Marketing 27 (2010) 91–106
Contents lists available at ScienceDirect
Intern. J. of Research in Marketing
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j r e s m a r
Innovation diffusion and new product growth models: A critical review and
research directions
Renana Peres a,b, Eitan Muller c,d,⁎, Vijay Mahajan e
a
School of Business Administration, Hebrew University of Jerusalem, Jerusalem, 91905, Israel
The Wharton School, University of Pennsylvania, Philadelphia, PA 19104, United States
Stern School of Business, New York University, New York, NY 10012, United States
d
Recanati Graduate School of Business Administration, Tel Aviv University, Tel Aviv, 69978, Israel
e
McCombs School of Business, University of Texas at Austin, Austin, TX 78712, United States
b
c
a r t i c l e
i n f o
Article history:
First received in 7, April 2009 and was under
review for 4 ½ months
Area editor: Marnik G. Dekimpe
Keywords:
Diffusion
Innovations
New product
Review
Social network
Individual level modeling
Agent based modeling
Multinational diffusion
Competition
Brand
Network externalities
Takeoff
Saddle
Technology generations
a b s t r a c t
Diffusion processes of new products and services have become increasingly complex and multifaceted in
recent years. Consumers today are exposed to a wide range of influences that include word-of-mouth
communications, network externalities, and social signals. Diffusion modeling, the research field in
marketing that seeks to understand the spread of innovations throughout their life cycle, has adapted to
describe and model these influences.
We discuss efforts to model these influences between and across markets and brands. In the context of a
single market, we focus on social networks, network externalities, takeoffs and saddles, and technology
generations. In the context of cross-markets and brands, we discuss cross-country influences, differences in
growth across countries, and effects of competition on growth.
On the basis of our review, we suggest that the diffusion framework, if it is to remain a state-of-the-art
paradigm for market evolution, must broaden in scope from focusing on interpersonal communications to
encompass the following definition: Innovation diffusion is the process of the market penetration of new
products and services that is driven by social influences, which include all interdependencies among consumers
that affect various market players with or without their explicit knowledge.
Although diffusion modeling has been researched extensively for the past 40 years, we believe that this field
of study has much more to offer in terms of describing and incorporating current market trends, which
include the opening up of markets in emerging economies, web-based services, online social networks, and
complex product–service structures.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
At the end of 2008, 4 billion people around the world were using
mobile phones (ITU, 2008; The Economist, 2009). Launched in 1981 in
Scandinavia, mobile phone service has become a part of everyday life
for more than half of the world's population residing in 211 countries.
Moreover, in several developed nations, the mobile phone has
reached a penetration level that now exceeds 100%, with consumers
adopting more than one handset, more than one phone number, and
possibly more than one provider. The massive penetration of mobile
telephony is not exceptional — many commonly used products and
services, such as DVDs, personal computers, digital cameras, online
banking, and the Internet, were unknown to consumers three decades
⁎ Corresponding author. Recanati Graduate School of Business Administration, Tel
Aviv University, Tel Aviv, 69978, Israel.
E-mail addresses: [email protected] (R. Peres), [email protected]
(E. Muller), [email protected] (V. Mahajan).
0167-8116/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijresmar.2009.12.012
ago. As firms invest continually in innovation, this influx of new
products and services is expected to continue into the future.
The spread of an innovation in a market is termed “diffusion”. Diffusion
research seeks to understand the spread of innovations by modeling their
entire life cycle from the perspective of communications and consumer
interactions. Traditionally, the main thread of diffusion models has been
based on the framework developed by Bass (1969). The Bass model
considers the aggregate first-purchase growth of a category of a durable
good introduced into a market with potential m. The social network into
which it diffuses is assumed to be fully connected and homogenous. At
each point in time, new adopters join the market as a result of two types of
influences: external influences (p), such as advertising and other
communications initiated by the firm, and internal market influences
(q) that result from interactions among adopters and potential adopters in
the social system. The Bass model states that the probability that an
individual will adopt the innovation — given that the individual has not
yet adopted it — is linear with respect to the number of previous adopters.
The model parameters p, q, and m can be estimated from the actual
92
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
adoption data. Parameter estimation issues are discussed in Jiang, Bass and
Bass (2006); Boswijk and Franses (2005); Van den Bulte and Stremersch
(2004); Venkatesan, Krishnan and Kumar (2004); Lilien et al., 2000;
Sultan, Farley and Lehmann (1990); and Van den Bulte and Lilien (1997).
The proliferation of newly introduced information, entertainment,
and communication products and services and the development of
market trends such as globalization and increased competition have
resulted in diffusion processes that go beyond the classical scenario of
a single market monopoly of durable goods in a homogenous, fully
connected social system. The diffusion modeling literature since 1990
has attempted to extend the Bass framework to reflect the increasing
complexity of new product growth. Table 1 provides an overview of
the main changes in research focus over the past two decades.
One of the fascinating shifts of focus described in Table 1 is an in-depth
discussion of the various types of internal influences involved in the
diffusion process. In the original article by Bass, as well as in many of the
diffusion studies that followed it, the internal parameter q was interpreted
as representing the influence of word of mouth between individuals.
Recent contributions to the diffusion modeling literature have reexamined this interpretation to identify and discuss other types of social
interactions. On the basis of these recent developments, we believe that
the definition of diffusion theory should be revised. The traditional
perception of diffusion as a theory of interpersonal communication
(Mahajan, Muller & Bass, 1990; Mahajan, Muller & Wind, 2000) should be
extended to encompass social interdependence of all kinds (Goldenberg
et al., 2010; Van den Bulte & Lilien, 2001). We therefore define diffusion of
innovation as follows:
Innovation diffusion is the process of the market penetration of new
products and services, which is driven by social influences. Such
influences include all of the interdependencies among consumers that
affect various market players with or without their explicit knowledge.
We discuss two types of additional social influences (besides
word-of-mouth communications) that have garnered recent interest:
network externalities and social signals.
Network externalities exist when the utility of a product to a
consumer increases as more consumers adopt the new product
(Rohlfs, 2001). Network externalities are considered to be direct if
utility is directly affected by the number of other users of the same
product, as in the case of telecommunication products and services
such as fax, phone, and e-mail. Network externalities can also be
indirect if the utility increases with the number of users of another,
complementary product. Thus, for example, the utility to a consumer
of adopting a DVD player increases with the increased penetration of
DVD titles (Stremersch & Binken, 2009; Stremersch, Tellis, Franses &
Binken, 2007). Interpersonal communication is not necessarily
needed for network externalities to work. Potential adopters can
find out about the penetration level of a new product from the media
or simply by observing retail offerings. For example, during the
Table 1
Shifts in focus of research interests.
Previous focus
Complemented with current focus
Word of mouth as driver
Monotonically increasing penetration
curve
Temporal
Industry-level analysis
Aggregate or segment-based models
Fully connected networks
Consumer interdependencies as drivers
Turning points and irregularities in the
penetration curve
Spatial
Brand-level analysis
Individual-level models
Partially connected and small-world
networks
Services
Managerial diagnostics
Products
Forecasting
transition from videotape to DVD, a consumer had merely to walk into
a Blockbuster movie rental store and observe the amount of aisle
space devoted to VHS vs. DVD to understand that DVDs were about to
become the new standard. We elaborate on network externalities in
Section 2.2.
Social signals relate to the social information that individuals infer
from adoption of an innovation by others. Through their purchases,
individuals may signal either social differences or group identity
(Bourdieu, 1984). These signals are transmitted to other individuals,
who follow the consumption behavior of people in their aspiration
groups (Simmel, 1957; Van den Bulte & Joshi, 2007; Van den Bulte &
Wuyts, 2007). Social signals operate vertically and horizontally. A
vertical social signal indicates the status of the adopter. Recent
research indicates that the competition for status is an important
growth driver, sometimes more important than interpersonal ties,
and that the speed of diffusion increases in societies that are more
sensitive to status differences (Van den Bulte & Stremersch, 2004).
Social signals are also transmitted horizontally to indicate group
identity. Adoption of an innovation by people in a given group signals
to members of that group to adopt and to members of other groups
who want to differentiate to avoid adoption (Berger & Heath, 2007;
2008). While social signals can be transmitted via word of mouth and/
or advertising, neither is a necessity. These signals are observed by
potential adopters who infer from them the social consequences of
adoption.
We note that a distinction should be made between social signals and
other types of signals, such as functional signals. Functional signals contain
information regarding the market perception of the functional attributes
of a product, such as its quality or the amount of risk involved in adopting
it, whereas social signals contain information regarding the social
consequences of adopting the product, including the social risk of
adopting the innovation. An important question is whether inclusion of
social inference and network externalities as internal influences contradicts the Bass framework. Traditional applications of the Bass framework
have interpreted internal influence in terms of word-of-mouth and
personal communications (Mahajan et al., 1990). However, this interpretation is not dictated by the model itself, which does not specify the drivers
of social contagion. Thus, the consumer interactions of network
externalities and social inference certainly fit the framework, as do
other possible growth drivers, as long as they imply that the probability of
purchase increases with the number of previous adopters.
In spite of growing evidence of the importance of personal
communication in product adoption, an alternative research branch has
emerged. This branch argues that the major driver of growth of new
products is consumer heterogeneity rather than consumer interaction.
The heterogeneity approach claims that the social system is heterogeneous in innovativeness, price sensitivity and needs, leading to heterogeneity in propensity to adopt. Thus, innovators are the least patient in
adopting, whereas laggards are the most patient. In such models, patience
is often inversely related to product affordability, consumer willingness to
pay, or reservation price (Bemmaor, 1994; Golder & Tellis, 1998; Russell,
1980; Song & Chintagunta, 2003). The dynamics of market volume are
determined by the shape of the distribution of “patience” in the face of
falling prices. If incomes are log-normally distributed in the population,
then growth is S-shaped (Golder & Tellis, 1997). This line of research
implies that the current approach of diffusion-based research has
overemphasized the influence of word-of-mouth communication (Van
den Bulte & Lilien, 2001, and Van den Bulte & Stremersch, 2004). Fig. 1
illustrates the range of possible drivers of new product diffusion, arranged
according to the level of direct interpersonal communication they involve.
Our objective in this paper is to review the interaction-based diffusion
literature published in the past decade and analyze how it has broadened
its scope to describe the richness of consumers' internal influences so as to
bring these influences in a unified way into the diffusion framework. We
do not aim in this paper to cover the entire diffusion literature; for that, we
refer the reader to recently published diffusion overviews (Mahajan et al.
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
93
2.1. Diffusion in social networks
Fig. 1. Market factors that drive the diffusion of innovations.
(2000); Meade and Islam (2006); Hauser, Tellis and Griffin (2006);
Chandrasekaran and Tellis (2007); and Krishnan and Suman (2009)).
Rather, we feel that there is a need for a review paper that integrates the
modeling efforts of various types of interpersonal influences into a single
framework and reviews studies that have explored the manifestations of
these influences within and across markets and brands.
We start by discussing consumer influences within a single market
in Section 2. We discuss issues such as modeling the social network,
network externalities, takeoffs and saddles, and technology generations. Consumer influences across markets and brands are discussed
in Section 3, where we relate to cross-country influences, differences
in growth across countries, and effects of competition on growth. In
Section 4, we suggest topics for further research. Fig. 2 illustrates the
section flow. Table 2 conveys a summary of the focus of the main
research efforts in each literature stream, as well as the corresponding
directions for further research.
2. Diffusion within markets and technologies
In this section, we discuss four of the seven most influential
diffusion-related areas studied in the past decade. These areas — social
networks, network externalities, takeoffs and saddles, and technology
generations — concern effects within a single market or technology.
The social network, or social system, is the substrate onto which an
innovation propagates. The gradual decrease in the efficiency of offline advertising (e.g., Evans, 2009) and the development of online
social networks such as Facebook have led firms to approach directly
the social networks of customers in their target markets and to invest
marketing efforts in enhancing the internal influences in those
networks. In order to successfully enhance such influences, a better
understanding is required as to how the structure and dynamics of the
social network influence the diffusion process.
A fundamental question regarding diffusion within social networks is how the social network structure influences product growth.
This question has not yet been answered theoretically but has been
explored in some empirical studies (see Van den Bulte & Wuyts, 2007,
for an overview). Much of the research attention has been on the roles
of central individuals (influentials, social hubs) on the overall growth
process (e.g., Goldenberg, Han, Lehmann & Hong, 2009; Iyengar et al.,
in press). Research on the role of network structure in diffusion is still
in its infancy, mainly because of the lack of data. New methods that
enable large-scale sampling and analyses of online networks will
probably boost further research (see, for example, Dorogovtsev &
Mendes, 2003; Jackson, 2008). From the modeling perspective, the
basic question is how to incorporate the social network into the
diffusion model. The implicit assumption of the Bass model and of
most of its extensions is that the social system is homogenous and
fully connected. Therefore, the adoption process can be appropriately
represented at the aggregate level. Aggregate diffusion models have
advantages as well as downsides (Parker, 1994): On one hand, they
are parsimonious and require few data for parameter estimation and
forecasting; on the other hand, they provide little intuition as to how
individual market interactions are linked to global market behavior.
Since the extensive recent research on social networks has revealed
that they are neither homogenous nor fully connected (Kossinets &
Watts, 2006), diffusion research is gradually extending its focus from
the aggregate level to an individual-level perspective.
One well-known approach for describing individual adoption
decisions and tying them to aggregate outcomes is agent-based
modeling, which describes the market as a collection of individual
elements (units, agents, or nodes) interacting with each other
through connections (links). The behavior (in our case, adoption) of
Fig. 2. The paper section flow.
94
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
Table 2
Research focus and future research directions.
Subject
Diffusion within markets Diffusion in social
and technologies
networks
Diffusion and
network
externalities
Takeoffs and
saddles
Technology
generations
Diffusion across markets Cross-country
and brands
influences
Growth differences
across countries
Competition and
growth
Research focus
Directions for further research
The roles of central individuals such as influentials
and experts
Individual-level modeling using agent-based models
and social network concepts
Incorporating findings on individual decision-making from behavioral
studies and choice experiments into diffusion modeling
The influence of social network characteristics (e.g., clustering) on
diffusion patterns
Measuring the effects of word-of-mouth campaigns
Explicit representations of word of mouth and signals
Use of data and tools from social network researchers
Network externalities in partially connected social networks
Types of externalities (direct vs. indirect, local vs.
global)
Incorporating externalities into the diffusion model
The impeding and enhancing effects of externalities
on the speed of diffusion
Defining takeoffs and saddles
Determinants and international comparisons of
takeoffs
Measurements of the size and frequency of saddles
Incorporating saddles into the diffusion framework
Acceleration of diffusion parameters across
generations
Multi-generational diffusion models
Multi-national diffusion models
Estimation of the magnitude of cross-country
influences
Optimal global entry strategies
Differences in diffusion parameters across countries
and their cultural and economic sources
Models of competitive diffusion including category
and brand level influences
The effects of disadoption and churn
each individual element is determined by a decision rule. Neural
networks, cellular automata, and small-world models are examples of
agent-based modeling techniques. A typical agent-based model is the
cellular automata of Goldenberg, Libai and Muller (2001a;2001b).
Each unit represents an individual consumer and has a value of “0” if it
has not yet adopted the product and “1” if it has. Potential adopters
(“units”) adopt due to the combination of external influences
(parameter p) and internal influences (parameter q) in a manner
similar to the Bass framework.
Such a model overcomes some of the limitations of aggregate-level
diffusion models. First, by establishing a connection between
individual-level influences and aggregate effects, it enables the
researcher to better relate individual-level marketing activities to
firm performance, which is measured at the aggregate level. The
response to marketing activities can be applied in the transition rule
of the individual agents, and the performance is the aggregation of the
decisions over the entire network. Second, this modeling approach
enables one to distinguish between interdependencies: for example,
Goldenberg et al. (2010) used it to explore network externalities by
adding a threshold to the decision rule. Third, this model allows for
heterogeneity by enabling individual susceptibility to influences (pi
and qi) to differ among units or by setting different link structures for
each unit. The model can be adapted to include almost any aspect of
heterogeneity, including individual responsiveness to price and
advertising (Libai, Muller & Peres, 2005), presence of negative word
of mouth (Goldenberg, Libai, Muller & Moldovan, 2007), intrinsic
consumer innovativeness (Goldenberg, Libai & Muller, 2002), presence of heavy users and connectors (Kumar, Petersen & Leone, 2007),
and individual roles in the social network — that is, hubs, connectors,
and experts (Goldenberg et al., 2009).
Another advantage of agent-based models is their ability to take
into account the spatial aspect of diffusion. Using a variation of agentbased models called small world, Goldenberg et al. (2001a) studied
spatial issues in the market through the relative influences of strong
Understanding the network externalities that bring about a tipping
point
Combining pre- and post-takeoff growth in the diffusion framework
Understanding the roles of heterogeneity and internal influences in
takeoff and saddle formation
Combining behavioral and modeling research to understand
technological substitution
Optimal timing for the release of a new generation
Advanced methods to improve forecasting accuracy
Diffusion of innovations in the developing world
Differences between western and emerging economies
Effects of demographic changes (e.g., immigration waves) on diffusion
The interaction between individual brand choice processes and
diffusion (1-step vs. 2-step process)
Influence of competition on the distribution chain and the effect on
diffusion
and weak ties. They showed that, consistent with Wuyts, Stremersch,
Van den Bulte and Franses (2004) and Rindfleisch and Moorman
(2001), the cumulative influence of weak ties has a strong effect on
the growth process. Garber, Goldenberg, Libai and Muller (2004)
suggested a measure of the spatial density of adoption to predict a
product's success or failure.
Conceptually, aggregate diffusion models represent the overall
results of individual-level processes. Therefore, in order to create a
coherent market representation, the individual-level and aggregate
market formulations should be equivalent. However, the equivalence
of the two formulations is not straightforward. Previous studies have
proposed methods of aggregating individual-level behavior based on
assumptions regarding customer heterogeneity and calculation of the
time to adoption (Chatterjee & Eliashberg, 1990; Van den Bulte &
Stremersch, 2004). In the specific case of cellular automata, Goldenberg et al. (2001a) demonstrated this equivalence by relating the
parameters p and q to the adoption hazard function and presenting
simulations that showed that the individual-level probability of
adoption generates diffusion curves with p and q. The relationship
between the Bass model and agent-based models was also investigated by Rahmandad and Streman (2008) and by Fibich et al. (2009).
Shaikh et al. (2006) showed how product adoption by units in a smallworld network can be aggregated to create the Bass model with some
relatively simple assumptions. However, the interface between the
individual level and the aggregate level still lacks a closed formulation
and needs further exploration.
2.2. Diffusion and network externalities
The dynamics of network externalities have received considerable
attention in the past two decades (see Stremersch et al., 2007). There
is as yet no consensus about the effect of network externalities on
growth rate. Conventional wisdom suggests that network effects drive
faster market growth due to the increasing returns associated with
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
such processes (Nair, Chintagunta & Dubé, 2004; Tellis, Yin & Niraj,
2009). However, networks can also create the opposite effect, slowing
growth with what is labeled “excess inertia” (Srinivasan, Lilien &
Rangaswamy, 2004). Early in the product life cycle, most consumers
see little utility in the product as there are few adopters; they may
take a “wait-and-see” approach until there are more adopters. Hence,
diffusion early on may be very slow and occur primarily among the
few consumers who see utility in the product despite a lack of
adoption by others. Overall, the process can be characterized by a
combination of excess inertia and excess momentum, i.e., slow
growth followed by a surge (Van den Bulte & Stremersch, 2006).
This growth pattern can occur via various types of network
externalities. In the case of direct network effects, which apply to fax, email, and other communication products, the number of adopters
drives utility directly because the utility of the product increases with
the number of adopters. (A fax machine is not useful if almost no one
else has one.) Regarding indirect network effects, as in the case of
hardware and software products, an increase in utility may occur
through market mediation (e.g., amount of compatible software
applications), which in turn is a function of the number of adopters.
Consumers generally wait to adopt hardware until there is enough
software to make it worthwhile. In addition, in the case of competing
standards, early adopters take the risk of choosing a standard that will
eventually “lose” and become obsolete, so many consumers wait until
it is clear which is the winning standard and, more importantly, which
standard or platform will no longer be supported.
The impact of network externalities on growth rate can be
determined by the source of the externalities under examination,
namely, global or local. Under global externalities, a consumer takes
into account an entire social system when evaluating the utility of a
product in terms of the number of adopters, whereas under local
externalities, a consumer considers adoption in relation to his close
social network. Research is gradually moving from considering only
global externalities towards exploring local externalities (Binken &
Stremersch, 2009). The marketing decisions of the firm can influence
the types of externalities that are relevant: growth of competing
standards will probably invoke a global effect since the “verdict” on
what eventually becomes the winning standard depends on the total
number of users. In contrast, a family program of a mobile phone
service provider might evoke a local effect since it involves only the
local social system.
The challenge in integrating network externalities into diffusion
models stems from the multiple effects of previous adopters on the
rate of growth. Previous users are expected to accelerate growth due
to interpersonal effects, including word of mouth and imitation,
which typically reduce both risk and search costs. Yet, the mere
adoption of previous adopters increases network externalities,
consequently enhancing growth. The literature on the modeling of
diffusion of innovations generally does not separate the two factors,
and a single parameter for internal influence is used to capture the
effects of both interpersonal communications and network externalities (Van den Bulte & Stremersch, 2004). Goldenberg et al. (2010)
separated the two effects by drawing on the literature on collective
action to relate network effects to threshold levels of adoption by
individual consumers. They found that network externalities have a
“chilling” effect on initial growth, which is then followed by a surge of
enhanced diffusion. These findings are further discussed in Gatignon
(2010), Rust (2010) and Tellis (2010).
2.3. Takeoffs and saddles
In the past decade, a stream of literature has emerged that
examines turning points in the product life cycle that are not included
in the classic smooth-adoption curve (see illustration in Fig. 3). We
focus here on research dealing with two turning points in the product
life cycle: takeoff, which occurs at the beginning, and saddle, which
95
Fig. 3. Turning points in the product life cycle.
occurs during early growth. The classic Bass model starts with
spontaneous adoption by an initial group of adopters but does not
provide explanations for the mechanisms that lead to this initial
adoption, or takeoff. Studies on takeoff focus on this initial stage and
explore the market's behavior and the interface between adoption
and the start of communication interactions.
Golder and Tellis (1997) defined takeoff time as the time at which a
dramatic increase in sales occurs that distinguishes the cutoff point
between the introduction and growth stage of the product lifecycle. The
importance of takeoff time to the firm is quite clear: a rapid increase in
sales requires substantial investments in production, distribution, and
marketing, which most often involve considerable lead time to put
into place successfully. Golder and Tellis (1997) applied a proportional hazard model to data that included 31 successful innovative
product categories in the U.S. between 1898 (automobiles) and 1990
(direct broadcast satellite media). They found that the average time to
takeoff for categories introduced after World War II was six years and
that average penetration at takeoff was 1.7% of market potential. Price
at takeoff was found to be 63% of the original price. Other studies have
investigated factors that influence time to takeoff. Accelerating factors
are price reduction, product category (brown goods such as CDs and
television sets take off faster than white goods), and cultural factors
such as masculinity and a low level of uncertainty avoidance (Foster,
Golder & Tellis, 2004; Tellis, Stremersch & Yin, 2003).
Takeoff as such does not require any consumer interaction.
Instead, it results from heterogeneity in price sensitivity and risk
avoidance—as the innovation's price is reduced, adoption becomes
associated with less risk, and the product takes off. Therefore, if one
believes that both heterogeneity and communication play a role in
new product adoption, takeoff is an excellent example of an interface
point: heterogeneity is dominant prior to takeoff, but consumer
interactions become the driving force immediately afterwards. As
already stated by Mahajan et al., 1990 review, there is a need for a
comprehensive theory that delves deeper into early market growth
prior to takeoff.
Following takeoff, diffusion models predict a monotonic increase
in sales up to the peak of growth. However, in some markets a sudden
decrease in sales may follow an initial rise. This decrease in sales was
observed by Moore (1991), who denoted it as the chasm between the
early and main markets, and this concept was later formalized and
explored by Mahajan and Muller (1998); Goldenberg et al. (2002);
Golder and Tellis (2004); Muller and Yogev (2006); Van den Bulte and
Joshi (2007); Vakratsas and Kolsarici (2008); and Libai, Mahajan and
Muller (2008). Goldenberg et al. (2002) referred to the phenomenon
as a “saddle” and defined it as a pattern in which an initial peak
predates a trough of a substantial depth and duration that is followed
by increased sales that eventually exceed the initial peak.
While a saddle can be attributed to causes such as changes in
technology and macroeconomic events, it can also be explained by
consumer interactions. Golder and Tellis (2004), as well as Chandrasekaran and Tellis (2006), have claimed that the saddle phenomenon can be explained using the informational cascade theory. Small
shocks to the economic system such as a minor recession can
96
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
temporarily decrease the adoption rate, and the decrease is magnified
through the informational cascade. Another explanation is based on
heterogeneity in the adopting population and its division into two
distinct groups. If these two groups adopt the innovation at widely
differing rates and have weak communication between them, sales
may show an interim trough (Goldenberg et al., 2002; Muller & Yogev,
2006; Van den Bulte & Joshi, 2007). These findings demonstrate how
combining heterogeneity and consumer interactions explains a
phenomenon that does not fit the typical bell-shaped sales curve.
2.4. Technology generations
In theory, the basic diffusion process is terminated by a decay of
the number of new adopters and saturation of the market potential. In
practice, however, products are often substituted with newer
generations of products with more advanced attributes. New product
growth across technology generations has garnered growing interest
among marketing scholars such as Bass and Bass (2001, 2004),
Mahajan and Muller (1996), and Norton and Bass (1987, 1992). A
major issue examined by these researchers is whether diffusion
accelerates between technology generations. This question has
practical importance for forecasting as projections regarding the
growth of advanced generations of a product must often be made
during the early stages of product penetration or before launch and
are thus based on using diffusion parameters from previous generations. Theoretically, this question is important, because it deals with
dependency within a sequence of diffusion processes and, more
broadly, with rigidity of the social system across generations. Does the
social system learn to improve its adoption skills across generations or
does it start each diffusion process from scratch? If it has learning
capabilities, how strong and how category-specific are they?
As Stremersch et al. (2010) pointed out, the literature offers
contradicting answers to the question of whether diffusion accelerates across technology generations. The key finding (or assumption)
of several studies across multiple product categories is that growth
parameters are constant across technology generations (Bass & Bass,
2004; Kim, Chang & Shocker, 2000; Mahajan & Muller, 1996; Norton &
Bass, 1992, 1987). Bayus (1994), for example, used a proportional
hazard model to analyze the diffusion of four generations of personal
computers and concluded that the average product lifespan did not
decline over time. This was true even when moderating variables
(such as year of entry and technology used) were included.
Exceptions to this premise were provided by Islam and Meade
(1997) and Pae and Lehmann (2003), who demonstrated that the
results can be an artifact of the difference in the length of time covered
by the data series used in the analysis; their findings were subject to
criticism by Van den Bulte (2004).
In a contradiction to the stability of growth parameters across
generations, there is a great body of evidence suggesting that the
overall temporal pattern of diffusion of innovation accelerates over
time (Van den Bulte & Stremersch, 2004;2006; Kohli, Lehmann & Pae,
1999). A recent analysis by Van den Bulte (2000;2002) found
conclusive evidence that such acceleration does indeed occur. Van
den Bulte investigated the issue of acceleration by adopting the Bass
model with the internal influence parameter (p) set to zero and
running the model on 31 product categories in consumer electronics
and household products. The average annual acceleration between
1946 and 1980 was found to be around 2%. Exceptions to this
generalized finding are rare (Bayus, 1994) and contested on the
grounds of estimation bias and invalid inference (Van den Bulte,
2000).
These two research streams form an intriguing paradox: It seems
that, in the same economy, an acceleration of the diffusion of
innovations over time should be reflected in an acceleration of
diffusion of technology generations that succeed one other; however,
the diffusion rates of sequential technology generations remain
constant. A resolution to the paradox was suggested recently by
Stremersch et al. (2010), who noted constant growth parameters
across generations but a shorter time to takeoff for each successive
generation. They investigated whether the faster takeoff of successive
generations is due to the passage of time or to the generational effect.
They defined technology generation as a set of products similar in
customer-perceived characteristics and technology vintage as the year
in which the first model of a specific technology generation was
launched commercially. Using a discrete proportional hazard model in
12 product categories, Stremersch, Muller, and Peres found that
acceleration in time to takeoff is due to the passage of time and not to
generational shifts. Thus, time indeed accelerates early growth,
whereas generational shifts do not.
The issue of technological substitution has raised questions related
to heterogeneity in the adopting population. Goldenberg and Oreg
(2007) proposed a redefinition of the “laggards” concept; they
suggested that laggards from previous product generations may
often become innovators of the latest generation because of
leapfrogging. Thus, for example, in the early days of the MP3
revolution, an early adopter of MP3 could be a user of a Walkman
cassette player who did not adopt CD technology and decided to
upgrade by leapfrogging to an MP3 player. Hence, early adopters of
MP3 players were not necessarily innovative; some may have been
leapfrogging laggards from previous generations. Thus, firms should
approach them with the appropriate marketing mix tools and not
treat them as innovators.
The entry of a new technology generation complicates the growth
dynamics and generates consumer-related processes that are not
observed in single-generation diffusion. First, the entry of a new
generation is usually considered to increase the market potential. In
addition, customers can upgrade and replace an old technology with a
new one. On the other hand, individuals who belong to the increased
market potential might decide eventually to adopt the older
generation of the product and, hence, cannibalize the new technology's potential. If there are more than two generations, adopters can
skip a generation and leapfrog to advanced versions. This means that
the entry of a new technology reveals heterogeneity in the adopting
population that was not realized in a single generation scenario.
Surprisingly, none of the diffusion models have yet offered a
comprehensive treatment of these dynamics. Studies so far have
focused on one or two of these aspects, such as upgrading (Bass &
Bass, 2001, 2004; Norton & Bass, 1992) and cannibalization (Mahajan
& Muller, 1996), but a unified theoretical treatment of the subject is
still required.
Normative decisions are also influenced by intergenerational
dynamics. Several papers have investigated optimal pricing decisions
under technological substitution (Padmanabhan & Bass, 1993;
Danaher, Hardie & Putsis, 2001; Lehmann & Esteban-Bravo, 2006).
However, with the exception of Lehmann and Esteban-Bravo (2006),
these studies did not address the dynamics of specific groups of
adopters.
3. Diffusion across markets and brands
In this section, we discuss the remaining three of the seven
research areas that, in our view, are the most significant in terms of
innovative diffusion research in the past decade. These three areas,
which concern various cross-market and cross-brand effects, are
cross-country influences, differences in growth across countries, and
effects of competition on growth.
3.1. Cross-country influences
A number of papers since 1990 have followed the global trend of
focusing on multinational product acceptance and have extended the
traditional single-market scope to explore problems and issues
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
related to international diffusion (see Dekimpe, Parker and Sarvary
(2000a) for a review).
A key issue in multinational diffusion, especially with respect to
order of entry, is the mutual influence of diffusion processes in various
countries. One of the major findings of the studies to date on crosscountry influences (with a few exceptions, such as Desiraju, Nair and
Chintagunta (2004) and Elberse and Eliashberg (2003)) is that the
entry time lag has a positive influence on the diffusion process; that is,
countries that introduce a given innovation later show a faster diffusion
process (Tellis et al., 2003; Dekimpe, Parker & Sarvary, 2000b;2000c;
Ganesh, Kumar & Subramaniam, 1997; Takada & Jain, 1991) and a shorter
time to takeoff (Van Everdingen, Fok & Stremersch, 2009). This crosscountry influence has been called the lead-lag effect; however, influence
can be multidirectional.
Several papers have modeled multi-market diffusion with crosscountry influences (Van Everdingen, Aghina & Fok, 2005; Kumar &
Krishnan, 2002; Ganesh et al., 1997; Putsis, Balasubramanian, Kaplan
& Sen, 1997; Eliashberg & Helsen, 1996; Ganesh & Kumar, 1996).
Generalizing these models, a generic cross-country influence model
may take the following form (xi is the proportion of adopters in
country i):
dxi ðtÞ
= ðpi + qi xi ðtÞ + ∑ δij xj ðtÞÞ⋅ð1−xi ðtÞÞ:
dt
j≠i
ð1Þ
The parameter δij represents cross-country effects between
country i and another country j. Cross-country effects can be a result
of two types of influence mechanisms: weak ties and signals. Weak
ties come from adopters in one country who communicate with
nonadopters from other countries (Wuyts et al., 2004; Rindfleisch &
Moorman, 2001). However, even without communicating with or
imitating other individuals, nonadopters are influenced by diffusion in
other countries. In other words, the level of acceptance of the
innovation in one country acts as a signal for customers in other
countries, reducing their perceptions of risk and increasing the
legitimacy of using the new product. While several studies stated
explicitly that the dominant effect was due to communication (Putsis
et al., 1997; Eliashberg & Helsen, 1996; Ganesh & Kumar, 1996),
others explored the effect without relating it to a specific mechanism
(e.g., Dekimpe et al., 2000b;2000c; Takada & Jain, 1991).
Note that although the distinction between weak ties and signals has
evident managerial implications, the commonly used aggregate models of
the type presented in Eq. (1) do not distinguish between the two effects;
both are represented through the parameters δij. Further research is
required to estimate the relative roles of word of mouth and noncommunication signals in cross-country spillover and to study their
relative effects on the overall diffusion process. Individual-level models
might be able to make this distinction.
Understanding cross-country influences is also valuable in the
context of normative managerial decisions in multinational markets.
Some studies have explored entry strategies — i.e., the question of
whether a firm should enter all of its markets simultaneously
(a “sprinkler” strategy) or sequentially (a “waterfall” strategy). Kalish,
Mahajan and Muller (1995) built a game-theoretic model for two
brands and suggested that the waterfall strategy is preferable when
conditions in foreign markets are unfavorable (slow growth or low
innovativeness), competitive pressure is low, the lead-lag effect is
high, and fixed entry costs are high. Libai et al. (2005) extended this
question to explore responsive budgeting strategies in which firms
dynamically allocate their marketing efforts according to developments in the market. Many other questions are still waiting to be
answered, and issues such as regulation (addressed by Stremersch &
Lemmens, 2009), international competition, and the optimal marketing mix of growing international markets can be further explored.
97
3.2. Growth differences across countries
Research on the evolution of multi-markets reveals a noteworthy
aspect of diffusion that deals with heterogeneity among different
social systems in which the same product is adopted. If various social
systems adopt the same product in different ways, understanding
these differences will clarify the diffusion process within each
country. A large number of studies published during the last two
decades have focused on describing and explaining such inter-country
differences. Key findings are summarized in Table 3. These studies
have generally focused on differences in the diffusion parameters p
and/or q (e.g., Van den Bulte, 2002; Dekimpe, Parker & Sarvary, 1998;
Helsen, Jedidi & DeSarbo, 1993; Takada & Jain, 1991), the ratio q/p
(Van den Bulte & Stremersch, 2004), time to takeoff (Tellis et al.,
2003) and duration of the growth stage (Stremersch & Tellis, 2004).
The salient result emerging from all of these papers is that
diffusion processes vary greatly among countries, even for the same
products or within the same continent (Ganesh, 1998; Mahajan &
Muller, 1994; Helsen et al., 1993). In addition to measuring the
differences among growth processes, these studies have also
investigated the country-specific sources of these differences. Besides
product entry time, discussed above, the underlying factors can be
divided into cultural sources and economic sources.
Cultural sources — These relate to the country's cultural characteristics and values. Takada and Jain (1991) found that the diffusion
parameter q is higher in countries that are high-context and
homophilous (such as Asian Pacific countries) relative to countries
such as the U.S. that are low-context and heterophilous. High-context
refers to a culture in which much of the information conveyed
through a communication resides in the context of the communication rather than in its explicit message, and homophilous implies that
communication takes place among individuals who share certain
characteristics. Similarly, Dekimpe et al. (2000b;2000c) that population heterogeneity has a negative effect on both time to adoption and
the probability of transition from nonadoption to partial or full
adoption in a country. Similar findings are described in Talukdar,
Sudhir and Ainslie (2002).
Dwyer, Mesak and Hsu (2005) used Hofstede (2001) dimensions of
national culture and found positive relationships between q and
collectivism (vs. individualism), masculinity (assertiveness and
competitiveness as desired traits), and high power distance (the
extent to which the less powerful members of institutions and
organizations in a country expect and accept that power is distributed
unequally). Their findings were supported by Van den Bulte and
Table 3
Sources of differences in diffusion parameters.
Source of difference
Entry time
Marketing
mix
Entry time lag
Price
Product type (brown
goods vs. white goods)
MarketExistence of competition
related
Regulation
Demographic Population heterogeneity
and cultural Population growth rate
No. of population centers
Hofstede's dimensions
Economic
Wealth (GDP, income per
capita)
Media availability
Income inequality
Regulation
Influence
Mostly positive
Mixed
Positive (brown goods: CD players,
TVs, camcorders, etc. penetrate faster)
Positive
Positive
Negative
Positive
Mixed
Positive
Direction: less uncertainly
avoidance, collectivism, masculinity,
power distance
Positive
Mixed
Positive
Negative
98
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
Stremersch (2004). In addition, Dwyer et al. (2005) found that shortterm orientation was positively associated with q, but they did not
observe a significant negative relationship between q and uncertainty
avoidance.
Economic sources — The influences of many macroeconomic
variables have been studied, yielding two main empirical generalizations: First, the wealth of the country (usually measured by gross
domestic product (GDP) per capita, but also by lifestyle, health status,
and urbanization) has a positive influence on diffusion (Desiraju et al.,
2004; Talukdar et al., 2002; Dekimpe et al., 2000b;2000c; Putsis et al.,
1997; Helsen et al., 1993). Note that wealth is not necessarily
equivalent to general welfare. For example, Van den Bulte and
Stremersch (2004) found a positive influence between the Gini index
for inequality and the ratio q/p. This finding is consistent with cultural
findings concerning positive associations with power distance and
masculinity. A second generalization is that access to mass media
(usually operationalized by the penetration of TV sets) has a positive
influence on the diffusion parameter p (Stremersch & Tellis, 2004;
Talukdar et al., 2002; Tellefsen & Takada, 1999; Putsis et al., 1997).
In pharmaceutical markets, regulation was found to influence
growth in sales (Stremersch & Lemmens, 2009). Tellis et al. (2003)
and Stremersch and Tellis (2004) distinguished between the
influences of cultural and economic factors on penetration stages.
They found, for example, that cultural factors influence time to
takeoff, whereas economic factors influence growth.
Some efforts have been made to include developing countries in
innovation diffusion studies, (e.g., Dekimpe et al., 1998;2000b,
Stremersch & Lemmens, 2009; Van Everdingen et al., 2009); however,
it remains to be determined whether emerging economies are
characterized by the patterns and forces that are at work in developed
economies or whether the theories have to be revised (Mahajan,
2009, Mahajan & Banga, 2006; Steenkamp & Burgess, 2002).
3.3. The effects of competition on growth
Competitive forces influence the growth of a new product and
decisions made about it. Although some innovative categories start as
monopolies, many quickly gain multiple competing brands. Interestingly, the existence of competition influences not only the interactions of customers with the firms, but also the dynamics of the
interactions of consumers with other consumers. Thus, a competitive
growing market reveals a range of consumer interdependencies that
are not present in the case of a monopoly: word of mouth now applies
to within-brand and cross-brand communications, compatibility
issues can increase or decrease the effect of network externalities,
and the entry of a new competitor can act as a signal of the quality of a
product.
Despite the richness of competitive phenomena in growing
markets, the early diffusion literature dealt with either growth of
monopolies or category-level growth. The marketing literature on
competition has investigated mainly mature markets, with some
studies of competitive effects in growing markets (see Chatterjee,
Eliashberg and Rao (2000) for a review). Only in the last two decades
have researchers combined these two research streams and incorporated some of the competitive effects into diffusion modeling.
Fig. 4 illustrates various competitive effects that modify and
influence the growth process and that do not exist in monopolies.
These competitive effects relate to customer flow and information
flow. Regarding customer flow, the firms compete on two fronts. The
first is acquisition of adopters from the market potential before their
competitors, and the second is prevention of customer churn and
acquisition of customers who churn from other brands. Competitors
might be legal brands or illegal brands; hence, a factor of coping with
piracy is also at play. The information flow under competition becomes
more complex since consumers' communication exists within and across
brands. Among these effects are the influences of competition on the
category growth rate and communication transfer within and between
brands.
At the category level, an intriguing empirical question is whether
competition enhances or delays category growth. Generally, competition has been found to have a positive effect on diffusion parameters
(Kauffman & Techatassanasoontorn, 2005; Van den Bulte & Stremersch, 2004; Kim, Bridges & Srivastava, 1999; Dekimpe et al., 1998).
An exception was observed by Dekimpe et al. (2000c), who showed
that an existing installed base of an old technology negatively affected
the growth of new technologies. Krishnan, Bass and Kumar (2000)
studied the impact of late entrants on the diffusion of incumbent
brands. Using data on diffusion of minivans and cellular phones in
several U.S. states, they found that the effect varied across markets. In
some markets, the market potential and the internal communication
parameter q increased with the entry of an additional brand, whereas
in other markets only one of these parameters increased. None of
these studies provides explanations of the mechanisms underlying
the acceleration. One potential explanation is that acceleration results
from greater marketing pressure on the target market. Kim et al.
(1999) have implicitly suggested that the number of competitors
constitutes a signal of the quality and long-term potential of the
product, which may result in acceleration. Alternatively, the positive
Fig. 4. Competitive effects on focal brand A of competition for market potential, piracy, cross-brand communication, and churning customers to and from the competitors.
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
effect may be a result of reduction in network externalities (Van den
Bulte & Stremersch, 2004). Clearly, further research is needed here.
Turning to the brand level, constructing a brand-growth model
requires discussing several conceptual issues. A basic issue is the
extent to which internal influence mechanisms operate at the brand
level. Some regard brand adoption as a two-stage process in which
consumers first adopt the category and then choose the brand (Givon,
Mahajan & Muller, 1995; Hahn, Park, Krishnamurti & Zoltners, 1994)
based on factors other than internal communication such as
promotion activities, price deals, and special offers. In spite of the
intuitive rationale of this approach, attempts to use it have been rare,
partially because it requires high-quality individual-level data. The
development of service markets and increased use of Customer
Relationship Management (CRM) systems by service providers can
facilitate data availability and promote the use of these types of
models. Landsman and Givon (2010) used banking data to investigate
the growth of financial products, and Weerahandi and Dalal (1992)
used business-to-business data for fax penetration. Such a two-stage
model is an excellent platform for incorporating customer heterogeneity (Lee, Lee & Lee, 2006; Jun & Park, 1999) into communicationbased approaches. The two-stage model enables an explicit investigation of the link between personal characteristics, product attributes,
individual choice behavior, and aggregate growth. Intensive research
is still needed to build firm theoretical and empirical infrastructures
for choice-based growth models.
Although the relationships between brand-level and categorylevel adoption have not yet been clearly identified, the main body of
literature has assumed that internal dynamics are important at the
brand level, and, therefore, a Bass-type model can be used to model
brand choice. Mapping the communications flow in the market, one
can say that a potential customer adopts brand i as a result of the
combination of two optional communication paths: within-brand
communication with adopters of brand i and/or cross-brand communication with adopters of other brands. Cross-brand communication
can influence a consumer's choice of a brand in two ways: the
consumer may receive negative information about the competing
brands or the consumer may receive information about the category
from adopters of other brands and subsequently adopt brand i
because its marketing mix is most appealing.
Generalizing from the competitive diffusion models published so
far, a diffusion equation for multiple brands that explicitly presents
both communication paths can take the following form (adopted from
Libai, Muller and Peres (2009a) and Savin and Terwiesch (2005)):
Nj ðtÞ
dNi ðtÞ
N ðtÞ
= ðpi + qi i
+ ∑ δij
Þðm−NðtÞÞ
dt
m
m
j≠i
ð2Þ
where N is the total number of adopters (N = Ni + Nj), and δij
represents the cross-brand influences. The parameter m is the overall
market potential. Many of the existing brand-level diffusion models
are special cases or variations of this generic model. Some assume that
within-brand communication equals cross-brand communication,
namely, δij = qi (Krishnan et al., 2000; Kim et al., 1999; Kalish et al.,
1995). Their underlying assumption is that there is no relevance to the
brand ownership of the individual who spreads the information.
Other modelers, such as Mahajan, Sharma and Buzzell (1993), assume
that all communications are brand-specific (δij = 0).
The relationship between the Bass model and the summation of
the equations of individual brands in each category should be further
studied. For example, when δij = qi , summing the equations for all
brands yields the category equation of the Bass model; however,
when cross-brand communication is not equal to within-brand
communication, the summation results in a model different from
the Bass model. Two studies have tried to examine systematically the
distinction between within- and cross-brand communication: Parker
and Gatignon (1994) and Libai et al. (2009a). For consumer goods
99
(Parker and Gatignon) and cellular services (Libai, Muller, and Peres),
the studies concluded that both within-brand and cross-brand
influences exist.
Note the formative resemblance between Eqs. (1) and (2): Both
describe consumer interdependencies; Eq. (1) represents influences of
consumers from other markets, whereas Eq. (2) deals with influences
from customers of competing firms. As argued above, δij can represent any
type of interdependency, and further research should be done to separate
the effects of word of mouth, signals, and network externalities. Empirical
and behavioral research is needed to elucidate what determines the ratio
of within-brand to cross-brand communications — it might be a
combination of the characteristics of the innovation, the social system,
and/or the nature of competition.
A conceptual difference between the two equations relates to
market potential. The multinational Eq. (1) describes diffusion
processes that operate in separate markets in which each firm
draws from its own market potential. In the competitive scenario of
Eq. (2), the assumption is that both firms compete for the same
market potential. Some studies have relaxed the assumption of a joint
potential and assumed that brands can develop independently, with
each brand having its own market potential (Parker & Gatignon,
1994). In that case, the market potential m in Eq. (2) should be
modified to mi, i.e., the equation should now be:
Nj ðtÞ
dNi ðtÞ
N ðtÞ
= ðpi + qi i
+ δij
Þðmi −Ni ðtÞÞ:
dt
mi
mj
ð3Þ
Note that Eq. (3) requires careful treatment and interpretation. If
one assumes that the market potentials of the two brands do not
overlap, then the brands do not compete for the attention and wallets
of the same potential consumers. On the other hand, if one assumes
that the market potentials of the brands do overlap and that the total
market potential m = ∑ mi, then this overall market potential
overestimates the true potential since the intersections should be
subtracted from the overall count. Mahajan et al. (1993) investigated
the market potential issue in the context of Polaroid's lawsuit against
Kodak, the latter having been accused of patent violation and of
attracting Polaroid's customers to a new brand of digital camera. By
breaking the nonadopter pool m − N into sub-pools according to the
market potential of each brand, the researchers concluded that Kodak
took about 30% of its customers from Polaroid's potential buyers.
However, at the same time, Kodak expanded the market for Polaroid
since about two thirds of Polaroid's sales would not have occurred if
Kodak had not entered the market.
Within- and cross-brand influences occur even among brands that
do not directly compete. Joshi, Reibstein and Zhang (2009) consider a
brand extension of a high-status market that develops a new, lowerstatus version of the product. The existing high-status market has a
positive influence on the new market, but the reciprocal social
influence of the new, low-status market on the old market is negative.
The example given is Porsche's entry into the SUV market: the target
customers of the Porsche SUV were metrosexual males, who were
negatively influenced by the profile of the existing adopters of the
category, the suburban “soccer moms”. Modeling of this phenomenon
was achieved by setting δ12and δ21 of Eq. (3) to be positive and
negative, respectively.
In addition to competing for market potential, firms can compete
for one another's existing customers for multi-purchase products such
as consumer goods and services. Brand switching, also termed
attrition, defection, or churn, is a major concern in many innovative
industries. Industry data imply that in the U.S. mobile phone industry,
the average annual attrition rate in 2005 was 26.2% (World Cellular
Information Service), while the average attrition rate for U.S.
companies in the 1990s was estimated at 20% (80% retention,
Reichheld, 1996). Attrition and its consequences have been discussed
in the CRM literature on mature markets. However, recent studies
100
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
have demonstrated that customer attrition can have a substantial
effect on growing markets (Gupta, Lehmann & Stuart, 2004; Hogan,
Lemon & Libai, 2003; Thompson & Sinha, 2008). Since most of the
studies in the diffusion literature deal with durable goods, researchers
have generally modeled the diffusion of services as if they were
durable goods and have not examined customer switching (Krishnan
et al., 2000; Lilien, Rangaswamy, and Van den Bulte, 2000; Jain,
Mahajan & Muller, 1991). Similarly, a few studies have attempted to
incorporate churn into the diffusion framework (Libai, Muller & Peres,
2009b; Givon, Mahajan & Muller, 1997; Hahn et al., 1994).
4. Directions for further research
From its inception, diffusion modeling has aimed to offer a
comprehensive description of the life cycle of innovative products.
In this paper, we have documented how technological developments
and changes in the nature of innovations have extended the scope of
classical diffusion questions. Future innovations are expected to
broaden this scope still further and reveal growth patterns not
previously observed. Taking the mobile phone industry example from
the introduction, we expect patterns such as long-term coexistence of
multiple technology generations, numerous services offered by
various (sometimes competing) providers on the same equipment,
and an increasing role of global considerations in the adoption
process. The enhanced penetration of communication and other
technological innovations in emerging economies, with their particular constraints and needs, provides a rich substrate for the
development of such patterns. To remain timely and abreast of
market trends, research in diffusion modeling will have to expand its
horizons. In this section we propose potential directions for that
expansion.
4.1. Diffusion, social networks, and network externalities: future
directions
The overall economic outcome of diffusion processes is usually
measured at the aggregate level. However, firms' marketing activities
often take place at the individual level. Such activities have increasingly
been aimed at influencing the internal dynamics of the market
(influential programs, buzz campaigns, etc.); maximizing their effectiveness requires a transition from an aggregate-level to an individuallevel perspective. We believe that this transition is a promising future
area of diffusion research.
Although modeling of individual adoption decisions started in the
1970s, exploration of those decisions in the context of a growing
market and through the lens of individual-level diffusion remains in
an early stage of development. This is mainly because it is difficult to
simultaneously map networks, collect individual-level data, and track
diffusion. However, the need for just such a study is increasing. The
online medium offers better opportunities for such research by
generating new types of individual-level data via blogs, CRM systems,
and sites like LinkedIn and Facebook.
To effectively investigate individual adoption decisions, researchers should elaborate on the individual-level models by separating the
adoption process into a hierarchy of effects (awareness, consideration,
liking, choice, purchase, and repeat purchase), integrating into each
stage findings from behavioral studies. The choice stage, for example,
has thus far been explored mainly through pre-purchase experiments,
such as conjoint experiments, but there are very few works (e.g.,
Landsman & Givon, 2010) that incorporate the choice stage into a
diffusion model. Another important stage in the hierarchy of effects is
the repeat purchase, which influences sales rather than adoption; this
is a major source of revenue in many service and goods industries.
Applying repeat-purchase individual-level mechanisms, as well as
developing models for sales rather than for adoption, can assist in
understanding the relative roles of repeat purchase vs. initial adoption
in the diffusion process and the influences of these factors on growth
and long-term profitability. Initial efforts were made by Prins, Verhoef
and Frances (2009) in the context of telecommunications services.
Individual-level models should also be extended to allow for
flexibility in determining the unit of adoption. Traditionally, that
unit has been the individual customer; however, adoption decisions
are not always made by a single individual. For example, for home
computing and communication products (cable TV; Wii vs. Sony
Playstation), the adoption decision is made by several household
members. Innovations in business markets have to be adopted by the
entire buying center and used by multiple individuals in the
organization. In models, such phenomena can be interpreted either
as network externalities or as examples of collective decision-making.
The structure of the social network is another factor that should be
taken into account when modeling individual-level decisions, because
it directly influences the speed and spatial pattern of diffusion and, as
a result, the marketing decisions of the firm. If, for example, the social
system is composed of isolated “islands” that communicate little with
one other, the firm should launch the product separately on each such
island in order to create global diffusion, whereas for other network
structures, the firm might be better off enhancing internal communications. Thus, researchers should delve deeper into the structure of
the specific social system involved and its influence on growth.
Another important characteristic of a social network, which can
considerably influence the diffusion pattern, is clustering, an element
that distinguishes social networks from random graphs. Clustering
means that if customer X is directly connected to Y and Y is directly
connected to Z, then there is an increased probability that X is also
directly connected to Z. There is little research so far that relates the
degree of clustering in a given social network to the speed of diffusion
of an innovation within that network. For example, Goldenberg et al.
(2001a) investigated the effects of weak ties on the speed of diffusion
but did not relate weak ties to clustering. Shaikh et al. (2006) studied
the role of clustering, but not on the speed of diffusion. Clustering is
expected to be associated with several effects: Within a cluster, the
speed of diffusion should increase with the level of clustering.
Between clusters, clustering should strengthen the role of weak ties
since information, once entered into a cluster with a high clustering
coefficient, will not likely leave the cluster unless pulled out by a weak
tie. Given the availability of agent-based models for theoretical
studies and network data for empirical studies, such issues appear to
be ripe for new research.
One of the notable marketing phenomena of recent times is firms'
attempts to impact their customers' word-of-mouth processes. Marketers
invest considerable resources in spreading information via word-ofmouth agent campaigns, referral reward programs, programs to identify
and impact influencers, online communities, viral marketing campaigns,
and a range of other programs (Godes & Mayzlin, 2009; De Bruyn & Lilien,
2008; Ryu & Feick, 2007; Dellarocas, 2003). A common means of word-ofmouth initiative is sampling, which is heavily used for cosmetics, food, and
online software distribution. In the case of software, a sample takes the
form of a free copy of the software, either with reduced features or for a
limited time. The complex nature of word-of-mouth dynamics, including
difficulties in following the spread of the effects of word of mouth and a
lack of established ways to measure its effects, makes the financial
justification of word-of-mouth programs a pressing issue, especially since
such initiatives compete for resources with traditional marketing efforts.
Using agent-based modeling along with empirical verification via actual
social networks, researchers are beginning to investigate approaches to
quantifying the effects of word-of-mouth programs (see Watts and Dodds
(2007) and Libai et al. (2009c)). One potential means of quantifying the
value of a member of such a program is to observe and measure that
person's ripple effect, i.e., the number of others that he or she affects
directly as well as second- and third-degree “infections”. Little work has
been done so far in terms of empirically measuring the effects of such
programs in general and in relation to social networks in particular.
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
An additional network-related issue that we believe should garner
more research attention is network externalities. The empirical
literature on network externalities, surprisingly, lacks evidence on
individuals' adoption threshold levels. For adoption to occur in the
presence of network externalities, a potential adopter has to overcome
two barriers. First, the consumer has to be convinced via the
communication process that the product provides good value. Second,
the consumer must be assured that the number of other adopters is such
that the network product will indeed supply its potential value, i.e., it
surpasses the consumer's individual threshold level. The shape of the
distribution of the thresholds within a population is of utmost
importance to the speed of diffusion. To see this, imagine a distribution
in which every person needs exactly one other individual who have
already adopted the product in order to adopt the product as well. Under
such a distribution, no one will adopt the product since all consumers
would be eternally waiting for someone else to adopt and initiate their
adoption process.
Given that social threshold modeling is already well grounded in the
sociological literature on collective action, one would imagine that the
issue of the distribution of thresholds is by now well established.
Unfortunately, this is not the case. We know of only a few (partial)
empirical verifications of Granovetter (1978) original claim that the
distribution is truncated normal: Ludemann (1999) and Goldenberg
et al. (2010). Taking on the network perspective requires disentangling
the effects of each of the three types of social influences — word-ofmouth, signals, and network externalities. In aggregate models such as
the Bass model, all three influences are embedded in the parameter q.
While some researchers have attempted to isolate these influences, more
knowledge is needed for managers to exercise some control over them.
For example, different types of word-of-mouth (e.g., within- and crossbrand, positive vs. negative) should be investigated; models should
separate various signals from word-of-mouth (e.g., weak ties vs. crosscountry signals, or cross-brand influences), and the signaling effects of
marketing mix variables should be studied. In addition, more empirical
studies are needed to find ways to identify, isolate, and measure the three
types of influences, their antecedents, and their influences on market
growth.
Network influences and network externalities have been extensively
researched in economics. One of the questions asked there concerns the
existence of a tipping point or critical mass necessary to jump-start the
growth process. However, the emphasis has been mainly on the state of
equilibrium rather than the dynamic path toward that steady state (see,
for example, Jackson (2008)). Most marketers will react with discomfort
at this notion as a characteristic of real markets. This is not to say that the
equilibrium concept is unimportant, but rather that it is incomplete. First,
the path to equilibrium is just as important, especially when the path
might take an inordinate amount of time. Second, because of the
Schumpeterian “creative destruction” process, in which an innovation
destructively changes economic and business environments, equilibrium
is never reached in most scenarios. On the path to equilibrium, an
innovation will disrupt the market, forming a new path to a new
equilibrium that also will never be reached.
The switch from aggregate-level modeling to analysis of diffusion
from a network perspective requires a revolution in data sources and
research tools. In addition to data on individual adoptions, information on social networks should be collected. For online networks,
automatic collection methods applied in computer science could be used
(Oestreicher-Singer & Sundararajan, 2008, 2009). For offline networks,
the task is more complicated. Some firms already insert connectivity
information into their CRM databases and survey their user groups and
loyalty program members regarding their social connections and other
people with whom they plan to share information about new products. In
other cases, researchers can collaborate with firms to collect such data.
The main tools of investigation will include network simulations such as
agent-based models, as well as methods that combine choice and
diffusion, such as Markov chains.
101
4.2. Life cycle issues: future directions
Two critical stages in the life cycle of a new product are its
beginning — until the takeoff point — and its substitution by a new
technology generation. The buildup to takeoff does not require any
consumer interaction; rather, takeoff is a result of heterogeneity in
price sensitivity and risk avoidance. Specifically, as the innovation price
decreases and becomes associated with less risk, the product takes off.
Therefore, if one believes that both heterogeneity and communication
play roles in new product adoption, then takeoff is an excellent example
of an interface point between these two factors: Heterogeneity is
dominant prior to takeoff, whereas consumer interactions become
dominant immediately afterwards. Understanding the communicationheterogeneity interplay has evident managerial importance. If, indeed,
the time to takeoff is controlled by risk and price perception, firms
would do better to invest in these aspects rather than to boost market
internal communication. However, takeoff studies so far have been
mostly descriptive and have not addressed the underlying mechanisms.
There is a need for a comprehensive theory and empirical analysis that
investigates this issue directly.
As for technological substitution, while models for the diffusion of
technology generations have existed for a while, major questions
remain unanswered. The first question relates to the substitution
process. According to traditional approaches, the new generation
eventually replaces the older generation; however, this is no longer
the case. For many products, old and new generations coexist for a
long time. In the mobile phone industry, the number of subscribers to
analog phones continued to increase long after digital technologies
became available. Use of older handset types in emerging economies
challenges manufacturers to cope simultaneously with multiple
technology generations. The two most frequently cited models of
technological substitution, developed by Norton and Bass (1987) and
Mahajan and Muller (1996), are restrictive in their treatment of the
coexistence of multiple generations. They also provide little insight
into other substitution issues, such as leapfrog behavior, and the
differences between adopter groups (e.g., new customers joining the
category vs. upgraders). Moreover, generational shift at the brand
level has not yet been tackled. We call for joint behavioral and
modeling research efforts to understand consumer behavior under
technological substitution. The findings of such studies should be
combined into a unified, comprehensive model.
A second question in the context of technological substitution relates
to the timing of the release of a new generation. The common wisdom
among practitioners is that the firm should introduce the product as
soon as it is available. This rule of thumb is supported by two main
studies in the field that indicate that the firm should introduce the new
generations either as soon as they are available or never (Wilson and
Norton, 1989) or at maturity of the old generation (Mahajan & Muller,
1996). Inclusion of factors such as cannibalization of market potentials
and competition among brands might alter these results. Studies in the
entertainment industry have investigated optimal timing of the release
of a new movie (Lehmann & Weinberg, 2000), but those models are
distinct from technological substitution of successive generations.
Normative studies regarding the optimal timing of generational entry
are called for.
The third question relates to forecasting the adoption of future
generations. A multi-generational category enables researchers to use
data from previous generations to forecast the diffusion of future
generations. In doing so, one might have to resort to semi-parametric
and non-parametric models (Stremersch & Lemmens, 2009; Sood,
James & Tellis, 2009).
4.3. Cross-country interactions and comparisons: future directions
Current demographic changes are affecting cross-country influences and raising new challenges for global marketers. Diffusion of
102
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
innovations in emerging economies is an increasingly important
managerial interest, especially in industries such as telecommunications, where market potential in the developed world is about to
approach its limit. The World Bank's report for 2008 indicates that,
while annual rates of growth in GDP in developed countries are
expected to remain around 2% in coming years, emerging economies
will continue to see stable and consistent growth of more than 7% per
year (The World Bank, 2008). Thus, emerging economies present
rapidly growing potential markets for innovations.
Diffusion of innovations in emerging economies involves unique
patterns that are rarely found in developed countries. For example,
while standard consumption in the mobile phone industry in the
developed world is one user per handset, in many emerging economies
the same mobile handset serves several family members, each with a
personal SIM card. Sometimes handsets, such as those given to family
members or household staff, are restricted so that they can only accept
incoming calls or call certain numbers (Chircu & Mahajan, 2009).
Consumers in the developed world frequently upgrade to new handsets,
whereas in developing regions, there is an active market for secondhand
handsets that are used with prepaid phone cards. The patterns of
consumption that are ubiquitous in the developing world can act as
springboards to larger questions, such as how prepaid services vs.
postpaid or package deals affect consumer choices and penetration or
how mobile phones can be used as financial intermediaries (Iyengar
et al., 2009; Mahajan, 2009; The Economist, 2009).
Although emerging economies usually fall behind developed
countries in the propensity to adopt innovations (The World Bank,
2008, p. 5), they are highly responsive to specific innovations that
meet their particular needs (Chircu & Mahajan, 2009). As an example,
executives of Comverse Technology recalled that early adopters of
their innovative product in the mid-1990s — voice mail — were
telephone service providers from Africa and India whose subscribers
used public phones and needed private voice mail boxes. Use of a
single handset by multiple individuals boosted the penetration of
prepaid phone cards in emerging economies and led service providers
to come up with creative mechanisms for separating billing for calls
from a single handset (Mahajan, 2009). To save air time, customers in
emerging economies make extensive use of ringtones to transfer
messages without answering the call; these customers use many
more ringtones and ringtone control functions than the average user
in a developed country. Despite the richness of the phenomena and
the growing availability of data, the main body of diffusion research
rarely relates to diffusion in emerging economies. The unique
diffusion patterns in those parts of the world have been cited as
anecdotal evidence but have not been incorporated into diffusion
models. With the exception of a few papers (e.g., Desiraju et al., 2004),
research has not addressed whether such patterns are countryspecific or abundant across emerging economies. We do not know
whether they are generated by growth drivers that are similar to
those already identified for developed countries or what actions firms
should take to maximize their profits under such growth patterns.
4.4. Competition and growth: future directions
Managerial decision-making usually applies to the firm's own
brands. As competitive structures become more complex, brand-level
decision-making becomes important in optimizing managerial decision-making. We have identified several research questions related to
competitive effects that would benefit from applying diffusion
modeling.
The first question relates to the scope of competition: Is there a single
market potential from which all brands draw, or is it a reasonable
working hypothesis that each brand has its own market potential? Since
having a distinct market from which to draw customers implies that
competitive pressures are relatively low, it seems that when competition is intense, the common market potential hypothesis is more
reasonable. However, models of both scenarios should be compared
empirically by applying them to a large set of data to resolve this
question. The answer to this question is important for determining the
true level of competitive intensity in diffusion markets and, consequently, for optimizing firms' decisions about marketing mix, branding
and positioning. For example, if market potentials do not overlap, firms
might direct fewer efforts to positioning against competition, be less
aggressive in their pricing, and be more willing to cooperate with their
competitors in distribution channels.
Second is the question of the influence of competition along the
distribution chain. In the mobile phone industry, for example, while
competing service providers distribute the same handset model, third
parties offer customers real time auto-selection of the network with
the best rate, so customers use the services of multiple service
providers. Besides the complex cross-firm dynamics involved, such
scenarios have important implications for customers' brand perceptions in such an environment and, in turn, for the brand strategy that
will optimize growth. Extending the basic diffusion model to include
both multiple layers and competition would improve descriptive and
normative investigations of this question.
Third is the still-open question regarding whether brand choice is
a one- or two-stage process. If brand choice is a two-stage process in
which consumer interactions are dominant in category choice and
special offers and advertising are dominant in choosing the brand, then
straightforward application of a standard diffusion model to brand-level
data is problematic. Although some insights into the brand choice
process derive from behavioral studies, diffusion modeling can combine
brand choice and individual-level decisions and estimate their relative
importance at each stage. Insights from such combined models might be
striking in terms of marketing mix decisions; they may aid in
establishing direct connections between product attributes, promotion
campaigns and brand elements and overall diffusion.
The fourth issue deals with the nature of consumer interactions
under competition. Take as an example the launch of the iPhone by
Apple, one of the most significant product introductions in the U.S. in
2007. This launch relied heavily on word-of-mouth communication and
buzz as opposed to paid advertising. While Apple was working under
the assumption that interpersonal communication would help push its
product, allowing a relatively limited investment in advertising, it
appears that others in the industry reaped the benefits of this crossbrand effect. To quote Verizon spokesperson Michael Murphy, “I would
have to think that a rising tide lifts all ships” (Reuters, 2007). The
distinction between consumer interactions at the category level and at
the brand level has received scant attention so far (Libai et al., 2009a)
but is crucial for managing the growth process. A better understanding
of such consumer interactions can assist in answering the following
questions: To what extent do operations of one brand influence the
diffusion of another? What are the implications for the amount and
types of advertising used? Which firms, in terms of size, market share,
and market potential, will benefit from the category- and brand-level
diffusion?
With the preceding example in mind, consider a social network in
which a customer spreads positive word of mouth about the iPhone.
One can compute the social value of this customer as the indirect
benefit that the firm receives not from his direct purchase, but rather
from his word-of-mouth activities. The social value stems from the
interplay of two sources: acquired customers and acceleration. Acquired
customers are customers whom the focal customer helps Apple to
acquire, customers whom Apple would not have acquired otherwise
because they would have been acquired by competitors. In addition, this
customer may cause a different set of customers to accelerate their
adoption of the iPhone: those who would have eventually chosen the
iPhone anyway. Untangling the effects of acquired customers and
acceleration on the growth of new products is necessary for quantifying
and measuring the social value of customers and the impact of word-ofmouth initiatives (see Libai et al. (2009c)).
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
A comprehensive brand-level perspective requires extension of
the types of product categories under research. The overwhelming
majority of the diffusion papers we have reviewed dealt with diffusion
of durable goods and entertainment products. In practice, many
products introduced to the market during the past few years have
been either services, such as digital cable TV or instant messaging, or
combined goods and services, such as mobile phones. Service-related
behaviors, such as attrition, multiple purchases, and ongoing word of
mouth, should be incorporated into diffusion modeling. First attempts
have been made by Gupta et al. (2004) and Libai et al. (2009b).
However, modeling should be directed more toward tying diffusion
concepts into the CRM literature to describe the influence of
relationship-related measures on the growth and valuation of
customers and firms.
Additional product-related modeling questions address interdependencies between products. For instance, with Amazon.com
offering more than 250,000 music albums as well as various choices
of book formats, portfolio decisions become crucial to a new product's
success. Decisions about the allocation of marketing efforts, such as
whether to focus on a limited number of blockbusters or to spread
efforts over many long-tailed products, are of great managerial
interest (Anderson, 2008). Extending the basic diffusion model to
include multi-product interactions can contribute to this discussion.
Another characteristic of growth of successful brands today, which
we expect to be enhanced in the future, is the existence of parallel
“shadow” diffusion processes. Givon et al. (1995) termed shadow
diffusion as a diffusion process that accompanies major diffusion growth
and influences it, yet is not captured in the adoption or sales data.
Although they used the term to describe piracy, we propose extending
its scope to describe a wider spectrum of latent parallel diffusion
processes that are part of neither sales nor adoption records. Shadow
diffusion is salient in the entertainment industry, where films, books,
and music CDs are advertised before they are launched so that
adoption decisions are made before the product is available (Hui,
Eliashberg & George, 2008). Although some appearances of shadow
diffusion have been discussed in the current literature, the subject
lacks thorough treatment. Future modeling research should describe
the variety of shadow processes and measure their influences on the
parameters and speed of the main diffusion process.
4.5. Practical applications in specific industries: future directions
A recurring call in previous reviews (Mahajan et al., 1990; Mahajan
et al., 2000), which we continue to emphasize here, has been for more
research and reports on the actual use of diffusion models in
marketing. Diffusion models are employed in two basic ways: to
develop a better general understanding of diffusion phenomena
(descriptive) and to predict diffusion paths for new technologies
(predictive) before there is a significant amount of data available.
Though such applications are common in practice — see, for example,
Ofek (2005) or Mauboussin (2004) — there is a need to clarify the
actual process of using the Bass framework and its extensions to
predict sales when data are not available. This applies particularly to
industries in which specialized diffusion models might be needed to
capture the idiosyncrasies of the industry. Two studies that have
demonstrated the issues and challenges of using the Bass framework
in real applications to forecast sales and adoption are Lilien,
Rangaswamy and Van den Bulte (2000) and Bass, Gordon, Ferguson
and Githens (2001). Three industries that seem especially of interest
to new diffusion modeling efforts are telecommunications, services,
and pharmaceuticals.
Telecom markets form a rich substrate of research opportunities.
In most cases these markets are well documented, and thus data are
relatively easy to obtain. Many market processes in telecom are
regulated; in certain scenarios, this regulation keeps market variables
under control, exposing hidden market mechanisms. Note that the
103
distribution structure of the telecom industry is complex: A single 3G
telephony end-user application depends on hardware and software
manufacturers, service providers, compatibility issues, and global
infrastructures, so specialized diffusion models might be called for.
Services are of special interest because of their wide availability
and the fact that they differ from durable goods in several important
aspects: First, many services involve multiple purchases. This might
not have a direct effect on the number of adopters, but it certainly has
an influence on the sales function. For many services, recurrent
purchasing leads to the development of long-term relationships
between customers and service providers; concepts and processes of
relationship marketing are therefore of relevance in these cases.
Though there have been few studies that model services, there is a
need to incorporate CRM into the diffusion framework for models of
innovation diffusion in the service sector.
In the pharmaceutical industry, a new wave of research recognizes
the unique nature of heath care management and especially the
idiosyncrasies of its supply chain. Thus, new approaches and models
have been formulated for modeling diffusion of pharmaceutical
products. See, for example, the special issue of this journal on
Marketing and Health (Stremersch, 2008; Vakratsas & Kolsarici, 2008;
Grewal, Chakravarty, Ding & Liechty, 2008). Solid pre-launch
forecasting methods are still needed, especially due to the pharmaceutical industry's high investment in R&D in very early stages of the
drug development process.
4.6. Afterthought
We have been writing review papers on research in innovation
diffusion with other researchers for the past thirty years (Mahajan &
Muller, 1979; Mahajan et al., 1990; Mahajan et al., 2000; Muller, Peres
and Mahajan, 2009). One could ask, what will the scope and content of
a review paper written in 2020 be? The answer to such a question
depends on the following:
• The development of new, complex types of product categories. Each
such new category will illuminate a different aspect of the diffusion
process and raise new research questions. Current examples are the
iPhone, which offers over 70,000 apps; the Kindle, with over
360,000 books; and cheap mobile phones with multiple SIM cards
for multiple users.
• The emergence of new modes of interdependencies between
consumers that connect their utility, consumption and communications patterns. Current examples are social network communities
such as Facebook, Amazon book recommendations and the
quintessential micro-blogging service Twitter.
• The continuing growth of data from social networking sites, mobile
phone service providers, e-mail communications, online communities and search engines such as Google Trends. Such data will open
numerous new possibilities for exploring individual adoption
decisions and linking them to overall diffusion patterns.
• The integration of cutting-edge modeling tools from other research
domains. Current examples are agent-based modeling and network
analysis as new tools for modeling growth of new products in
networked markets, and nonparametric regressions as forecasting
tools.
• The level of firms' intervention in the internal dynamics of the
market. The emergence of “amplified word of mouth,” in which
firms use tools such as seeding programs to affect diffusion, is
currently a highly popular market trend. Should this trend continue,
diffusion studies will pay more attention to optimization questions
on social networks. However, should this intervention prove
unprofitable or impractical or face legal and regulatory hurdles,
researchers and practitioners will pursue other influence
mechanisms.
104
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
The driving force behind diffusion research, as with any other
research field, is an enthusiastic community of inquisitive researchers
with a command of cutting-edge research tools, who inspire each
other in extending knowledge boundaries and pursuing intriguing
questions. We hope that this review has provided an overview of the
collective effort of this community in the last decade, as well as a
glimpse of what is yet to come.
Acknowledgements
The authors would like to thank the then editor, Stefan Stremersch,
the area editor and two anonymous reviewers for a number of
thoughtful suggestions and comments. The authors also greatly
benefited from the comments and suggestions on previous drafts by
Bart Bronnenberg, Deepa Chandrasekaran, Jacob Goldenberg, Towhidul Islam, Trichy Krishnan, Barak Libai, Philip Parker, Ashutosh Prasad,
Arvind Rangaswami, John Roberts, Gerard Tellis, Christophe Van den
Bulte, and Charles Weinberg.
References
Anderson, C. (2008). The long tail. New York: Hyperion.
Bass, F. M. (1969). A new product growth model for consumer durables. Management
Science, 15(1), 215−227.
Bass, P., & Bass, F. M. (2001). Diffusion of technology generations: A model of adoption
and repeat sales. Working paper: University of Texas at Dallas.
Bass, P., & Bass, F. M. (2004). IT waves: two completed generational diffusion models.
Working paper: University of Texas at Dallas.
Bass, F. M., Gordon, K., Ferguson, T. L., & Githens, M. L. (2001). DIRECTV: Forecasting
diffusion of a new technology prior to product launch. Interfaces, 31(3), S82−S93.
Bayus, B. L. (1994). Are product life cycles really getting shorter? Journal of Product
Innovation Management, 11(4), 300−308.
Bemmaor, A. C. (1994). Modeling the diffusion of new durable goods: Word-of-mouth
effect versus consumer heterogeneity. In G. Laurent, G. L. Lilien, & B. Pras (Eds.),
Research traditions in marketing (pp. 201−223). Boston: Kluwer.
Berger, J., & Heath, C. (2007). When consumers diverge from others. Identity signaling
and product domains. Journal of Consumer Research, 34(2), 121−134.
Berger, J., & Heath, C. (2008). Who drives divergence? Identity signaling, out-group
similarity, and the abandonment of cultural tastes. Journal of Personality and Social
Psychology, 95(3), 593−607.
Binken, J. L. G., & Stremersch, S. (2009). The effect of superstar software on hardware
sales in system markets. Journal of Marketing, 73(2), 88−104.
Boswijk, P. H., & Franses, P. H. (2005). On the econometrics of the Bass diffusion model.
Journal of Business & Economic Statistics, 23(3), 255−288.
Bourdieu, P. (1984). Distinction: A social critique of the judgment of taste. Cambridge, MA:
Harvard University Press.
Chandrasekaran, D., & Tellis, G. J. (2006). Getting a grip on the saddle: Cycles, chasms, or
cascades? PDMA Research Forum, Atlanta, 21–22 October.
Chandrasekaran, D., & Tellis, G. J. (2007). A critical review of marketing research on
diffusion of new products. In N. K. Malhorta (Ed.), Review of marketing research
(pp. 39−80). Armonk, NY: M. E. Sharpe.
Chatterjee, R., & Eliashberg, J. (1990). The innovation diffusion process in a heterogeneous
population: A micromodeling approach. Management Science, 36(9), 1057−1079.
Chatterjee, R., Eliashberg, J., & Rao, V. R. (2000). Dynamic models incorporating
competition. In V. Mahajan, E. Muller, & Y. Wind (Eds.), New-product diffusion
models. New York: Kluwer Academic Publisher.
Chircu, A. M., & Mahajan, V. (2009). Revisiting digital divide: An analysis of mobile
technology depth and service breadth in the BRIC countries. Journal of Product
Innovation Management, 26(4), 455−466.
Danaher, P. J., Hardie, B. G. S., & Putsis, W. P. (2001). Marketing-mix variables and the
diffusion of successive generations of technological innovation. International
Journal of Marketing Research, 38(4), 501−514.
De Bruyn, A., & Lilien, G. (2008). A multi-stage model of word-of-mouth influence through
viral marketing. International Journal of Research in Marketing, 25(3), 151−163.
Dekimpe, M. G., Parker, P. M., & Sarvary, M. (1998). Staged estimation of international
diffusion models: An application to global cellular telephone adoption. Technological Forecasting and Social Change, 57(1–2), 105−132.
Dekimpe, M. G., Parker, P. M., & Sarvary, M. (2000). Multimarket and global diffusion. In
V. Mahajan, E. Muller, & Y. Wind (Eds.), New-product diffusion models. New York:
Kluwer Academic Publisher.
Dekimpe, M. G., Parker, P. M., & Sarvary, M. (2000). Global diffusion of technological
innovations: A coupled-hazard approach. Journal of Marketing Research, 37(1), 47−59.
Dekimpe, M. G., Parker, P. M., & Sarvary, M. (2000). Globalization: Modeling technology
adoption timing across countries. Technological Forecasting and Social Change, 63
(1), 25−42.
Dellarocas, C. (2003). The digitization of word of mouth: Promise and challenges of
online feedback mechanisms. Management Science, 49(10), 1407−1424.
Desiraju, R., Nair, H., & Chintagunta, P. (2004). Diffusion of new pharmaceutical drugs in
developing and developed nations. International Journal of Research in Marketing, 21
(4), 341−357.
Dorogovtsev, S. N., & Mendes, J. F. (2003). Evolution of networks. Oxford: Oxford
University Press.
Dwyer, S., Mesak, H., & Hsu, M. (2005). An exploratory examination of the influence of
national culture on cross-national product diffusion. Journal of International
Marketing, 13(2), 1−27.
Elberse, A., & Eliashberg, J. (2003). Demand and supply dynamics for sequentially
released products in international markets: The case of motion pictures. Marketing
Science, 22(3), 329−354.
Eliashberg, J., & Helsen, K. (1996). Modeling lead / lag phenomena in global marketing:
the case of VCRs. Working paper: University of Pennsylvania.
Evans, D. S. (2009). The online advertising industry: Economics, evolution, and privacy.
Journal of Economic Perspectives, 23(3), 37−60.
Fibich, G., Gibori, R., & Muller, E. (2009). Analysis of cellular automata diffusion models
in marketing. Working paper: Tel Aviv University.
Foster, J. A., Golder, P. N., & Tellis, G. J. (2004). Predicting sales takeoff for Whirlpool's
new personal valet. Marketing Science, 23(2), 182−185.
Ganesh, J. (1998). Converging trends within the European Union: Insights from an
analysis of diffusion patterns. Journal of International Marketing, 6(4), 32−48.
Ganesh, J., & Kumar, V. (1996). Capturing the cross-national learning effect: An analysis
of industrial technology diffusion. Journal of the Academy of Marketing Science, 24
(4), 328−337.
Ganesh, J., Kumar, V., & Subramaniam, V. (1997). Learning effect in multinational
diffusion of consumer durables: An exploratory investigation. Journal of the
Academy of Marketing Science, 25(3), 214−228.
Garber, T., Goldenberg, J., Libai, B., & Muller, E. (2004). From density to destiny: Using
spatial dimension of sales data for early prediction of new product success.
Marketing Science, 23(3), 419−428.
Gatignon, H. (2010). Commentary on Jacob Goldenberg, Barak Libai, and Eitan Muller's
“The chilling effects of network externalities”. International Journal of Research in
Marketing, 27(1), 16−17.
Givon, M., Mahajan, V., & Muller, E. (1995). Software piracy: Estimation of lost sales and
the impact on software diffusion. Journal of Marketing, 59(1), 29−37.
Givon, M., Mahajan, V., & Muller, E. (1997). Assessing the relationship between the user-based
market share and unit sales-based market share for pirated software brands in
competitive markets. Technological Forecasting and Social Change, 55(2), 131−144.
Godes, D., & Mayzlin, D. (2009). Firm-created word-of-mouth communication:
Evidence from a field test. Marketing Science, 28(4), 721−739.
Goldenberg, J., & Oreg, S. (2007). Laggards in disguise: Resistance to adopt and the
leapfrogging effect. Technological Forecasting and Social Change, 74(8), 1272−1281.
Goldenberg, J., Libai, B., & Muller, E. (2001). Using complex systems analysis to advance
marketing theory development: Modeling heterogeneity effects on new product growth
through stochastic cellular automata. Academy of Marketing Science Review, 9.
Goldenberg, J., Libai, B., & Muller, E. (2001). Talk of the network: A complex system look
at the underlying process of word of mouth. Marketing Letters, 12(3), 211−223.
Goldenberg, J., Libai, B., & Muller, E. (2002). Riding the saddle: How cross-market
communications can create a major slump in sales. Journal of Marketing, 66(2), 1−16.
Goldenberg, J., Libai, B., Muller, E., & Moldovan, S. (2007). The NPV of bad news.
International Journal of Research in Marketing, 24(3), 186−200.
Goldenberg, J., Han, S., Lehmann, D. R., & Hong, J. W. (2009). The role of hubs in
adoption processes. Journal of Marketing, 73(2), 1−13.
Goldenberg, J., Libai, B., & Muller, E. (2010). The chilling effect of network externalities.
International Journal of Research in Marketing, 27(1), 4–15.
Golder, P. N., & Tellis, G. J. (1997). Will it ever fly? Modeling the takeoff of really new
consumer durables. Marketing Science, 16(3), 256−270.
Golder, P. N., & Tellis, G. J. (1998). Beyond diffusion: An affordability model of the
growth of new consumer durables. Journal of Forecasting, 17(3–4), 259−280.
Golder, P. N., & Tellis, G. J. (2004). Growing, growing, gone: Cascades, diffusion, and
turning points in the product life cycle. Marketing Science, 23(2), 207−218.
Granovetter, M. (1978). Threshold models of collective behavior. American Journal of
Sociology, 83(6), 1420−1443.
Grewal, R., Chakravarty, A., Ding, M., & Liechty, J. (2008). Counting chickens before the
eggs hatch: Associating new product development portfolios with shareholder
expectations in the pharmaceutical sector. International Journal of Research in
Marketing, 25(4), 261−272.
Gupta, S., Lehmann, D. R., & Stuart, J. A. (2004). Valuing customers. Journal of Marketing
Research, 41(1), 7−18.
Hahn, M., Park, S., Krishnamurti, L., & Zoltners, A. A. (1994). Analysis of new product
diffusion using a four-segment trial-repeat model. Marketing Science, 13(3),
224−247.
Hauser, J., Tellis, G. J., & Griffin, A. (2006). Research on innovation: A review and agenda
for Marketing Science. Marketing Science, 25(6), 687−717.
Helsen, K., Jedidi, K., & DeSarbo, W. S. (1993). A new approach to country segmentation
utilizing multinational diffusion patterns. Journal of Marketing, 57(4), 60−71.
Hofstede, Geert (2001). Culture's consequences. Thousand Oaks, CA: Sage Publications.
Hogan, J. E., Lemon, K. N., & Libai, B. (2003). What is the real value of a lost customer?
Journal of Service Research., 5(3), 196−208.
Hui, S. K., Eliashberg, J., & George, E. I. (2008). DVD preorder and sales: An optimal
stopping approach. Marketing Science, 27(6), 1097−1110.
Islam, T., & Meade, N. (1997). The diffusion of successive generations of a
technology: A more general model. Technological Forecasting and Social
Change, 56(1), 49−60.
ITU (International Telecommunication Union). (2008). World Information Society 2007
report. Available as www.itu.int/wisr.
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
Iyengar, R., Koenigsberg, O., & Muller, E. (2009). The effects of uncertainty on
mobile phone package size and subscribers choice. Working paper: Columbia
University.
Iyengar, R., Van den Bulte, C., & Valente, T. W. (in press). Opinion leadership and social
contagion in new product diffusion. Marketing Science.
Jackson, M. O. (2008). Social and economic networks. Princeton: Princeton University
Press.
Jain, D., Mahajan, V., & Muller, E. (1991). Innovation diffusion in the presence of supply
restrictions. Marketing Science, 10(1), 83−90.
Jiang, Z., Bass, F. M., & Bass, P. I. (2006). The virtual Bass model and the left-hand
truncation bias in diffusion of innovation studies. International Journal of Research in
Marketing, 23(1), 93−106.
Joshi, Y. V., Reibstein, D. J., & Zhang, J. Z. (2009). Optimal entry timing in markets with
social influence. Management Science, 55(6), 926−939.
Jun, D. B., & Park, Y. S. (1999). A choice-based diffusion model for multiple generations
of products. Technological Forecasting and Social Change, 61(1), 45−48.
Kalish, S., Mahajan, V., & Muller, E. (1995). Waterfall and sprinkler new-product
strategies in competitive global markets. International Journal of Research in
Marketing, 12(2), 105−119.
Kauffman, R. J., & Techatassanasoontorn, A. A. (2005). International diffusion of digital
mobile technology: A coupled-hazard state-based approach. Information Technology and Management, 6(2–3), 253−292.
Kim, N., Bridges, E., & Srivastava, R. K. (1999). A simultaneous model for innovative
product category sales diffusion and competitive dynamics. International Journal of
Research in Marketing, 16(2), 95−111.
Kim, N., Chang, D. R., & Shocker, A. D. (2000). Modeling intercategory and generational
dynamics for a growing information technology industry. Management Science, 46
(4), 496−512.
Kohli, R., Lehmann, D. R., & Pae, J. (1999). Extent and impact of incubation time in new
product diffusion. Journal of Product Innovation Management, 16(2), 134−144.
Kossinets, G., & Watts, D. J. (2006). Empirical analysis of an evolving social network.
Science, 311(5757), 88−90.
Krishnan, T. V., & Suman, A. T. (2009). International diffusion of new products. In M.
Kotabe, & K. Helsem (Eds.), The sage handbook of international marketing
(pp. 325−345). London: Sage Publication.
Krishnan, T. V., Bass, F. M., & Kumar, V. (2000). The impact of late entrant on the
diffusion of a new product/service. Journal of Marketing Research, 37(2), 269−278.
Kumar, V., & Krishnan, T. V. (2002). Research note: Multinational diffusion models: An
alternative framework. Marketing Science, 21(3), 318−332.
Kumar, V., Petersen, A. J., & Leone, R. P. (2007). How valuable is word of mouth? Harvard
Business Review, 85(10), 139−146.
Landsman, V., & Givon, M. (2010). The diffusion of a new service: Combining
service consideration and brand choice. Quantitative Marketing and Economics, 8(1),
91–121.
Lee, J. Y. C., Lee, J. -D., & Lee, C. -Y. (2006). Forecasting future demand for large-screen
television sets using conjoint analysis with diffusion model. Technological
Forecasting and Social Change, 73(4), 362.
Lehmann, D. R., & Esteban-Bravo, M. (2006). When giving some away makes sense to
jump-start the diffusion process. Marketing Letters, 17(4), 243−254.
Lehmann, D. R., & Weinberg, C. B. (2000). Sale through sequential distribution
channels: an application to movies and videos. Journal of Marketing, 64(3), 18−33.
Libai, B., Muller, E., & Peres, R. (2005). The role of seeding in multi-market entry.
International Journal of Research in Marketing, 22(4), 375−393.
Libai, B., Mahajan, V., & Muller, E. (2008). Can you see the chasm? Innovation diffusion
according to Rogers, Bass and Moore. In N. Malhorta (Ed.), Review of Marketing
Research. Armonk, NY: ME Sharpe Publications.
Libai, B., Muller, E., & Peres, R. (2009). The influence of within-brand and cross-brand word of
mouth on the growth of competitive markets. Journal of Marketing, 73(2), 19−34.
Libai, B., Muller, E., & Peres, R. (2009). The diffusion of services. Journal of Marketing
Research, 46(2), 163−175.
Libai, B., Muller, E., & Peres, R. (2009c). The social value of word-of-mouth programs:
Acceleration vs. Acquisition. Working paper: Tel Aviv University.
Lilien, G. L., Rangaswamy, A., & Van den Bulte, C. (2000). Diffusion models: Managerial
applications and software. In V. Mahajan, E. Muller, & Y. Wind (Eds.), New product
diffusion models. New York: Kluwer Academic Publishers.
Ludemann, C. (1999). Subjective expected utility, thresholds, and recycling. Environment and Behavior, 31(5), 613−629.
Mahajan, V. (2009). Africa rising. Philadelphia: Wharton School Publishing.
Mahajan, V., & Banga, K. (2006). The 86% solution: How to succeed in the biggest marketing
opportunity of the next 50 years. Philadelphia: Wharton School Publishing.
Mahajan, V., & Muller, E. (1979). Innovation diffusion and new product growth models
in marketing. Journal of Marketing, 43(4), 55−68.
Mahajan, V., & Muller, E. (1994). Will the 1992 unification of the European Community
accelerate diffusion of new ideas, products, and technologies? Technological
Forecasting and Social Change, 45(3), 221−235.
Mahajan, V., & Muller, E. (1996). Timing, diffusion, and substitution of successive
generations of technological innovations: The IBM mainframe case. Technological
Forecasting and Social Change, 51(2), 109−132.
Mahajan, V., & Muller, E. (1998). When is it worthwhile targeting the majority instead of the
innovators in a new product launch? Journal of Marketing Research, 35(3), 488−495.
Mahajan, V., Muller, E., & Bass, F. M. (1990). New product diffusion models in
marketing: A review and directions for research. Journal of Marketing, 54(1), 1−26.
Mahajan, V., Sharma, S., & Buzzell, R. D. (1993). Assessing the impact of competitive entry
on market expansion and incumbent sales. Journal of Marketing, 57(3), 39−52.
Mahajan, V., Muller, E., & Wind, J. (2000). New product diffusion models. New York:
Kluwer Academic Publishers.
105
Mauboussin, M. J. (2004). The economics of customer businesses: Legg Mason Capital
Management report. Available at. http://www.lmcm.com/pdf/TheEconomicsofCustomerBusinesses.pdf.
Meade, N., & Islam, T. (2006). Modeling and forecasting the diffusion of innovation — A
25 year review. International Journal of Forecasting, 22(3), 519−545.
Moore, G. A. (1991). Crossing the chasm. New York: HarperBusiness.
Muller, E., & Yogev, G. (2006). When does the majority become a majority? Empirical
analysis of the time at which main market adopters purchase the bulk of our sales.
Technological Forecasting and Social Change, 73(10), 1107−1120.
Nair, H., Chintagunta, P., & Dubé, J. -P. (2004). Empirical analysis if indirect network
effects in the market for personal digital assistants. Quantitative Marketing and
Economics, 2(1), 23−58.
Norton, J. A., & Bass, F. M. (1987). A diffusion theory model of adoption and substitution
for successive generations of high-technology products. Management Science, 33
(9), 1069−1086.
Norton, J. A., & Bass, F. M. (1992). The evolution of technological generations: The law of
capture. Sloan Management Review, 33(2), 66−77.
Oestreicher-Singer, G., & Sundararajan, A. (2008). The visible hand of social networks in
electronic markets. Working paper: New York University.
Oestreicher-Singer, G. & Sundararajan A. (2009). Recommendation Networks and the
Long Tail of Electronic Commerce. Working paper: New York University.
Ofek, E. (2005). Forecasting the adoption of new products. Harvard Business School
Case9–505–062.
Padmanabhan, V., & Bass, F. M. (1993). Optimal pricing of successive generations of
product advances. International Journal of Research in Marketing, 10(2), 185−207.
Pae, J. H., & Lehmann, D. R. (2003). Multigeneration innovation diffusion: The impact of
intergeneration time. Journal of the Academy of Marketing Science, 31(1), 36−45.
Parker, P. M. (1994). Aggregate diffusion forecasting models in marketing: A critical
review. International Journal of Forecasting, 10(2), 353−380.
Parker, P., & Gatignon, H. (1994). Specifying competitive effects in diffusion models: An
empirical analysis. International Journal of Research in Marketing, 11(1), 17−39.
Prins, R., Verhoef, P. C., & Frances, P. H. (2009). The impact of adoption timing on new
service usage and early disadoption. International Journal of Research in Marketing,
26(4), 304−313.
Putsis, W. P., Balasubramanian, S., Kaplan, E. H., & Sen, S. K. (1997). Mixing behavior in
cross-country diffusion. Marketing Science, 16(4), 354−370.
Rahmandad, H., & Streman, J. (2008). Heterogeneity and network structure in the
dynamics of diffusion: Comparing agent-based and differential equation models.
Management Science, 54(5), 998−1014.
Reichheld, F. F. (1996). The loyalty effect. Cambridge, MA: Harvard Business School
Press.
Reuters. (2007). “Apple Builds Hype for iPhone with Less”. June 20.
Rindfleisch, A., & Moorman, C. (2001). The acquisition and utilization of information in
new product alliances: A strength-of-ties perspective. Journal of Marketing, 65(2),
1−18.
Rohlfs, J. (2001). Bandwagon effects in high-technology industries. Cambridge, MA: MIT
Press.
Russell, T. (1980). Comments on the relationship between diffusion rates, experience
curves, and demand elasticities for consumer durable technological innovations.
Journal of Business, 53(3), 69−73.
Rust, R. (2010). Network externalities—not cool? A comment on ‘The chilling effects of
network externalities'. International Journal of Research in Marketing.
Ryu, G., & Feick, L. (2007). A penny for your thoughts: Referral reward programs and
referral likelihood. Journal of Marketing, 71(1), 84−94.
Savin, S., & Terwiesch, C. (2005). Optimal product launch times in a duopoly:
Balancing life-cycle revenues with product cost. Operations Research, 53(1),
26−47.
Shaikh, N. I., Rangaswamy, A., & Balakrishnan, A. (2006). Modeling the diffusion of
innovations using small-world networks. Working paper: Penn State University.
Simmel, G. (1957). Fashion. The American Journal of Sociology, 62(6), 541−558.
Song, I., & Chintagunta, P. K. (2003). A micromodel of new product adoption with
heterogeneous and forward-looking consumers. Quantitative Marketing and
Economics, 1(4), 371−407.
Sood, A., James, G. A., & Tellis, G. J. (2009). Functional regression: A new model for
predicting market penetration of new products. Marketing Science, 28(1), 36−51.
Srinivasan, R., Lilien, G. L., & Rangaswamy, A. (2004). First in, first out? The surprising
effects of network externalities on pioneer survival. Journal of Marketing, 68(1),
41−58.
Steenkamp, J. -B. E., & Burgess, S. M. (2002). Optimum stimulation level and exploratory
consumer behavior in an emerging consumer market. International Journal of
Research in Marketing, 19(2), 131−150.
Stremersch, S. (2008). Health and marketing: The emergence of a new field of research.
International Journal of Research in Marketing, 25(4), 229−233.
Stremersch, S., & Binken, J. L. G. (2009). The effect of superstar software on hardware
sales in system markets. Journal of Marketing, 73(2), 88−104.
Stremersch, S., & Lemmens, A. (2009). Sales growth of new pharmaceuticals across the
globe: The role of regulatory regimes. Marketing Science, 28(4), 690−708.
Stremersch, S., & Tellis, G. J. (2004). Understanding and managing international
growth of new products. International Journal of Research in Marketing, 21(4),
421−438.
Stremersch, S., Tellis, G. J., Franses, P. H., & Binken, J. L. G. (2007). Indirect network
effects in new product growth. Journal of Marketing, 71(3), 52−74.
Stremersch, S., Muller, E., & Peres, R. (2010). Does new product growth accelerate
across technology generations? Journal of Marketing, 21(2), 103−120.
Sultan, F., Farley, J. H., & Lehmann, D. R. (1990). A meta-analysis of applications of
diffusion models. Journal of Marketing Research, 27(1), 70−77.
106
R. Peres et al. / Intern. J. of Research in Marketing 27 (2010) 91–106
Takada, H., & Jain, D. (1991). Cross-national analysis of diffusion of consumer durable
goods in Pacific Rim countries. Journal of Marketing, 55(2), 48−54.
Talukdar, D., Sudhir, K., & Ainslie, A. (2002). Investigating new product diffusion across
products and countries. Marketing Science, 21(1), 97−116.
Tellefsen, T., & Takada, H. (1999). The relationship between mass media availability and
the multicountry diffusion of consumer products. Journal of International Marketing,
7(1), 77−96.
Tellis, G. J. (2010). Network effects: Do they warm or chill a budding product?
International Journal of Research in Marketing, 27(1), 20−21.
Tellis, G. J., Stremersch, S., & Yin, E. (2003). The international takeoff of new products:
the role of economics, culture, and country innovativeness. Marketing Science, 22
(2), 188−208.
Tellis, G. J., Yin, E., & Niraj, R. (2009). Does quality win? Network effects versus quality in
high-tech markets. Journal of Marketing Research, 46(2), 135−149.
The Economist. (2009). A special report on telecoms in emerging markets. September
26, 1-19.
The World Bank (2008). The World Bank Report, 2008. Global economic prospects:
Technology diffusion in the developing world. Washington, DC: The International
Bank for Reconstruction and Development / The World Bank.
Thompson, S. A., & Sinha, R. K. (2008). Brand communities and new product adoption:
The influence and limits of oppositional loyalty. Journal of Marketing, 72(6),
65−80.
Vakratsas, D., & Kolsarici, C. (2008). A dual-market diffusion model for a new prescription
pharmaceutical. International Journal of Research in Marketing, 25(4), 282−293.
Van den Bulte, C. (2000). New product diffusion acceleration: Measurement and
analysis. Marketing Science, 19(4), 366−380.
Van den Bulte, C. (2002). Want to know how diffusion speed varies across countries and
products? Try using a Bass model. PDMA Visions, 26, 12−15.
Van den Bulte, C. (2004). Multigeneration innovation diffusion and intergeneration
time: A cautionary note. Journal of the Academy of Marketing Science, 32(3),
357−360.
Van den Bulte, C., & Joshi, Y. V. (2007). New product diffusion with influentials and
imitators. Marketing Science, 26(3), 400−421.
Van den Bulte, C., & Lilien, G. L. (1997). Bias and systematic change in the parameter
estimates of macro-level diffusion models. Marketing Science, 16(4), 338−353.
Van den Bulte, C., & Lilien, G. L. (2001). Medical innovation revisited: Social contagion
versus marketing effort. American Journal of Sociology, 106(5), 1409−1435.
Van den Bulte, C., & Stremersch, S. (2004). Social contagion and income heterogeneity
in new product diffusion: A meta-analytic test. Marketing Science, 23(4), 530−544.
Van den Bulte, C., & Stremersch, S. (2006). Contrasting early and late new product
diffusion: speed across time, products, and countries. Working paper: Erasmus
University.
Van den Bulte, C., & Wuyts, S. (2007). Social networks and marketing, MSI relevant
knowledge series. Cambridge, MA: Marketing Science Institute.
Van Everdingen, Y. M., Aghina, W. B., & Fok, D. (2005). Forecasting cross-population
innovation diffusion: A Bayesian approach. International Journal of Research in
Marketing, 22(3), 293−308.
Van Everdingen, Y. M., Fok, D., & Stremersch, S. (2009). Modeling global spill-over in
new product takeoff. Journal of Marketing Research, 46(5), 637−652.
Venkatesan, R., Krishnan, T. V., & Kumar, V. (2004). Evolutionary estimation of macrolevel diffusion models using genetic algorithms: An alternative to nonlinear least
squares. Marketing Science, 23(3), 451–464.
Watts, D., & Dodds, P. S. (2007). Influentials, networks, and public opinion formation.
Journal of Consumer Research, 34(4), 441−458.
Weerahandi, S., & Dalal, S. R. (1992). A choice-based approach to the diffusion of a service:
Forecasting fax penetration by market segments. Marketing Science, 11(1), 39−53.
World Cellular Information Service. WCIS database, http://www.wcisdata.com/
Wilson, O. L., & Norton, J. A. (1989). Optimal entry timing for a product line extension.
Marketing Science, 8(1), 1–17.
Wuyts, S., Stremersch, S., Van den Bulte, C., & Franses, P. H. (2004). Vertical marketing
systems for complex products: A triadic perspective. Journal of Marketing Research,
41(4), 479−487.